Until quite recently, the neutral beams used in magnetic fusion research, material processing, etching, sterilization and other applications were all formed from positive ions. Positive hydrogen isotope ions were extracted and accelerated from gas discharge plasma by electrostatic fields. Immediately after the ground plane of the accelerator, they entered a gas cell, where they underwent both charge exchange reactions to acquire an electron and impact ionization reactions to lose it again. Because the charge exchange cross section falls much more rapidly with increasing energy than does the ionization cross section, the equilibrium neutral fraction in a thick gas cell begins to drop rapidly at energies greater than 60 keV for hydrogen particles. For hydrogen isotope neutral beam applications requiring energies appreciably higher than this, it is necessary to produce and accelerate negative ions, and to then convert them to neutrals in a thin gas cell, which can result in a neutral fraction of about 60% across a wide range of energies up to many MeVs. Even higher neutral fractions can be obtained if a plasma or photon cell is used to convert energetic negative ion beams to neutrals. In the case of a photon cell, for which photon energy exceeds electron affinity of hydrogen, neutral fractions could be close to 100%. It is worthwhile to note that the first time the idea of the application of negative ions in accelerator physics was stated by Alvarez more than 50 years ago [1].
Since neutral beams for current drive and heating on larger fusion devices of the future, as well as some applications on present-day devices, require energies well beyond that accessible with positive ions, negative-ion-based neutral beams were developed in recent years. However, beam currents achieved so far are significantly less than that produced quite routinely by positive ion sources. A physical reason for the lower performance of negative ion sources in terms of beam current is the low electron affinity of hydrogen, which is only 0.75 eV. Therefore, it is much more difficult to produce negative hydrogen ions than their positive counterparts. It is also quite difficult for newly born negative ions to reach an extraction region without collisions with energetic electrons which, with very high probability, will cause the loss of the extra loosely bound electron. Extracting H− ions from plasma to form a beam is likewise more complicated than with H+ ions, since the negative ions will be accompanied by a much larger current of electrons unless suppression measures are employed. Since the cross section for collisional stripping of the electron from an H− ion to produce an atom is considerably greater than the cross section for an H+ ion to acquire an electron from a hydrogen molecule, the fraction of ions converted to neutrals during acceleration can be significant unless the gas line density in the accelerator path is minimized by operating the ion source at a low pressure. Ions prematurely neutralized during acceleration form a low energy tail, and generally have greater divergence than those which experience the full acceleration potential.
Neutralization of the accelerated negative ion beam can be done in a gas target with an efficiency of about 60%. The usage of plasma and photon targets allows for the further increase in the neutralization efficiency of negative ions. Overall energy efficiency of the injector can be increased by recuperation of the energy of the ion species remaining in the beam after passing a neutralizer.
The schematic diagram of a high-power neutral beam injector for the ITER tokomak, which is also typical for other reactor-grade magnetic plasma confinement systems under consideration, is shown in FIG. 3 [2]. The basic components of the injector are a high-current source of negative ions, an ion accelerator, a neutralizer, and a magnetic separator of the charged component of the charge-exchanged beam with ion collectors-recuperators.
In order to sustain the required vacuum conditions in the injector, a high vacuum pumping system typically is used with large size gate valves cutting the beam duct from the plasma device and/or providing access to major elements of the injector. The beam parameters are measured by using retractable calorimetric targets, as well as by non-invasive optical methods. Production of powerful neutral beams requires a corresponding power supply to be used.
According to the principle of production, the sources of negative ions can be divided into the following groups:                volume production (plasma) sources—in which ions are produced in the volume of plasma;        surface production sources—in which ions are produced on the surface of electrodes or special targets;        surface-plasma sources—in which ions are produced on the surfaces of electrodes interacting with plasma particles, which were developed by the Novosibirsk group [3]; and        charge-exchange sources—in which negative ions are produced due to the charge-exchange of the accelerated positive ion beams on different targets.        
To generate plasma in modern volume H− ion sources similar to that in the positive ion source, arc discharges with hot filaments or hollow cathodes are used, as well as RF discharges in hydrogen. For the improvement of electron confinement in the discharge and for the decrease of the hydrogen density in the gas-discharge chamber, which is important for negative ion sources, discharges in a magnetic field are used. The systems with an external magnetic field (i.e., with Penning or magnetron geometry of electrodes, with electron oscillation in the longitudinal magnetic field of the “reflective” discharge), and the systems with a peripheral magnetic field (multipole) are widely used. A cutaway view of the discharge chamber with a peripheral magnetic field developed for the neutral beam injector of JET is shown in FIG. 4 [3]. A magnetic field at the periphery of the plasma box is produced by permanent magnets installed on its outer surface. The magnets are arranged in rows in which magnetization direction is constant or changes in staggered order, so that magnetic field lines have geometry of linear or checkerboard cusps near the wall.
Application of the systems with a multipole magnetic field at the periphery of the plasma chambers in particular, allows the systems to maintain a dense plasma in the source at the reduced gas working pressure in the chamber down to 1-4 Pa (without cesium) and down to 0.3 Pa—in the systems with cesium [4]. Such a reduction of hydrogen density in the discharge chamber is particularly important for high current multi-aperture giant ion sources which are being developed for applications in fusion research.
At the moment, surface plasma production ion sources are considered the most suitable for production of high current negative ion beams.
In surface plasma production ion sources the ions are produced in interaction between particles having sufficient energy and a low work function surface. This effect can be enhanced by alkali coating of the surface exposed to the bombardment. There are two principal processes, namely the thermodynamic-equilibrium surface ionization, where the slow atom or molecule impinging on the surface is emitted back as a positive or negative ion after a mean residence time, and the non-equilibrium (kinetic) atom-surface interaction, where negative ions are produced by sputtering, impact desorption (in contrast to thermal desorption where the thermal particles are desorbed) or reflection in the presence of an alkali metal coating. In the process of the thermodynamic-equilibrium ionization the adsorbed particles come off the surface in the conditions of thermal equilibrium. The ionization coefficient of the particles leaving the surface is determined by the Saha formula and appears to be very small ˜0.02%.
The process of non-equilibrium kinetic surface ionization appears to be much more effective in the surface and has a low enough work function comparable to electron affinity of the negative ion. During this process, the negative ion comes off the surface overcoming the near surface barrier using kinetic energy acquired from the primary particle. Near the surface an energy level of the additional electron is lower than the upper Fermi level of the electrons in metal and this level can be very easily occupied by electron tunneling from metal. During ion movement off the surface it overcomes a potential barrier produced by image charge
      U    image    =      -                            e          2                          4          ⁢          x                    .      The field of the charge image heightens the energy level of the additional electron relative to the energy levels of the electrons in metal. Starting from some critical distance, the level of the additional electron becomes higher than the upper energy level of the electrons in the metal, and resonance tunneling returns back the electron from the leaving ion back to the metal. In case the particle is coming off fast enough, the coefficient of negative ionization appears to be quite high for the surface with low work function which can be provided by covering an alkali metal, especially cesium.
It is experimentally shown that the degree of negative ionization of hydrogen particles coming off this surface with a lowered work function may reach
      β    -    =                    j        -                              j          -                +                  j          0                +                  j          +                      =          0      ,      67.      It is noted that the work function on tungsten surfaces has a minimum value with Cs coverage of 0.6 monolayers (on a tungsten crystal 110 surface).
For the development of negative hydrogen ion sources, it is important that the integral yield of negative ions is sufficiently high, K−=9-25%, for collisions of hydrogen atoms and positive ions with energies of 3-25 eV with surfaces with low work function, like Mo+Cs, W+Cs [5]. In particular, (see FIG. 5) in the bombardment of a cesiated molybdenum surface by Frank-Condon atoms with energy greater than 2 eV, the integral conversion efficiency into H− ions may reach K−˜8%.
In surface-plasma sources (SPSs) [3], the negative ion production is realized due to kinetic surface ionization—processes of sputtering, desorption or reflection on electrodes in contact with the gas-discharge plasma. The electrodes of special emitters with a lowered work function are used in SPSs for the enhancement of negative ion production. As a rule, the addition of a small amount of cesium into the discharge allows one to obtain a manifold increase in the luminosity and intensity of H− beams. Cesium seeding into the discharge remarkably decreases the accompanying flux of electrons extracted with the negative ions.
In an SPS, gas discharge plasma serves several functions, namely it produces intense fluxes of particles bombarding the electrodes; the plasma sheath adjacent to the electrode produces ion acceleration, thereby increasing the energy of the bombarding particles; negative ions, which are produced at electrodes under negative potential, are accelerated by the plasma sheath potential and come through the plasma layer into the extraction region without considerable destruction. An intense negative ion production with rather high power and gas efficiencies was obtained in various modifications of SPS under “dirty” gas-discharge conditions and an intense bombardment of the electrodes.
Several SPS sources have been developed for large fusion devices like LHD, JT-60U and the international (ITER) tokomak.
Typical features of these sources can be understood considering the injector of a LHD stellarator [4], which is shown in FIG. 6 [4, 6]. Arc plasma is produced in a large magnetic multipole bucket fence chamber with a volume of ˜100 Liters. Twenty four tungsten filaments support the 3 kA, ˜80 V arc under hydrogen pressure of about 0.3-0.4 Pa. An external magnet filter with a maximal field at center of ˜50 G provides the electron density and temperature decrease in the extraction region near the plasma electrode. Positive bias of plasma electrode (˜10 V) decreases an accompanying electron flux. Negative ions are produced on the plasma electrode covered by optimal cesium layer. External cesium ovens (three for one source) equipped with pneumatic valves supply the distributed cesium seeding. Negative ion production attains a maximum at optimal plasma electrode temperature of 200-250° C. The plasma electrode is thermally insulated and its temperature is determined by power loads plasma discharge.
A four electrode multi-aperture ion-optical system, which is used in the LHD ion source, is shown in FIG. 7 [6]. Negative ions are extracted through 770 emission apertures with a diameter of 1.4 cm each. The apertures occupy an area of 25×125 cm2 on the plasma electrode. Small permanent magnets are embedded into the extraction grid between apertures to deflect the co-extracted electrons from the beam onto the extraction electrode wall. An additional electron suppression grid, installed behind the extraction grid suppressed the secondary electrons, backscattered or emitted from the extracted electrode walls. A multi-slit grounded grid with high transparency is used in the ion source. It decreases the beam intersection area thus improving the voltage holding capacity and lowering the gas pressure in the gaps by a factor of 2.5 with the corresponding reduction of the beam stripping losses. Both the extraction electrode and the grounded electrode are water-cooled.
Cesium seeding into the multi-cusp source provides a 5-fold increase of an extracted negative ion current and a linear growth of H− ions yield in the wide range of discharge powers and hydrogen filling pressures. Other important advantages of cesium seeding are a ˜10-fold decrease of the co-extracted electron current and an essential decrease of hydrogen pressure in the discharge down to 0.3 Pa.
The multi-cusp sources at LHD routinely provide about a 30 A ion current each with current density of 30 mA/cm2 in 2 second long pulses [6]. The main issues for the LHD ion sources is a blocking of cesium, which is seeded to the arc chamber, by the tungsten sputtered from filaments and the decrease of high voltage holding capacity when operated in the high-power long pulse regime.
The negative-ion-based neutral beam injector of the LHD has two ion sources operated with hydrogen at nominal beam energy of 180 keV. Every injector has achieved the nominal injection power of 5 MW during 128 sec pulse, so that each ion source provides a 2.5 MW neutral beam. FIGS. 8A and B shows the LHD neutral beam injector. A focal length of the ion source is 13 m, and the pivot point of the two sources is located 15.4 m downstream. Injection port is about 3 m long with the narrowest part being 52 cm in diameter and 68 cm in length.
The ion sources with RF plasma drivers and negative ion production on a plasma electrode covered by cesium are under development at IPP Garching. The RF drivers produce more clean plasma, so that there is no cesium blocking by tungsten in these sources. Steady state extraction of a negative ion beam pulse with a beam current of 1 A, energy of ˜20 kV and duration of 3600 seconds was demonstrated by IPP in 2011.
At present, high energy neutral beam injectors, which are under development for next phase fusion devices, such as, e.g., the ITER tokomak, have not demonstrated stable operation at a desired 1 MeV energy and steady state or continuous wave (CW) operation with high enough current. Therefore, there is a need to develop viable solutions whenever it is possible to resolve the problems preventing achievement of the target parameters of the beam, such as, e.g., beam energy in the range of 500-1000 KeV, effective current density in neutrals of the main vessel port of 100-200 A/m3, power per neutral beam injector of about 5-20 MW pulse length of 1000 seconds, and gas loads introduced by the beam injector to be less than 1-2% of the beam current. It is noted that achievement of this goal becomes much less demanding if a negative ion current in a module of the injector is reduced down to a 8-10 A extracting ion current compared to a 40 A extracting ion current for the ITER beam. The stepping down in the extracted current and beam power would result in strong alterations in the design of the key elements of the injector ion source and the high energy accelerator, so that much more well developed technologies and approaches become applicable improving the reliability of the injector. Therefore, present consideration suggests the extracted current of 8-10 A per module, under assumption that the required output injection power can be obtained using several injector modules producing high current density, low divergent beams.
The surface plasma source performance is rather well documented and several ion sources now in operation have produced continuous scalable ion beams in excess of 1 A or higher. So far, key parameters of neutral beam injectors, like beam power and pulse duration, are quite far from those required for the injector under consideration. Current status of the development of these injectors can be understood from Table 1.
TABLE 1TAEITERJT-60ULHDIPPCEA-JAERICurrent density200 D−100 D−350 H−230 D−216 D−(A/m2)280 H−330 H−195 H−Beam energy1000 H−1000 D− 365186925(keV)100 H−Pulse length (s)≧10003600 D− 1910<65 3 H−1000Electron to ion ratio1  ~0.25<1<1<1pressure (Pa)0.30.30.260.30.30.35commentsCombined numbers notFilamentFilamentRF source, not fullKamabokoIIIyet achieved, experimentssourcesourceextraction, test bedsource (JAERI)under way at IPP Garching -known as BATMANon MANTIS (CEA)long pulse sourceoperated at 2 A/20 kVMANITU now deliversfor ~6 s1 A/20 kV for up to3600 s with D−
Therefore, it is desirable to provide an improved neutral beam injector.